Exploiting the proteomics revolution in biotechnology: from disease and antibody targets to optimizing bioprocess development

Abstract

Recent advancements in proteomics have enabled the generation of high-quality data sets useful for applications ranging from target and monoclonal antibody (mAB) discovery to bioprocess optimization. Comparative proteomics approaches have recently been used to identify novel disease targets in oncology and other disease conditions. Proteomics has also been applied as a new avenue for mAb discovery. Finally, CHO and Escherichia coli cells represent the dominant production hosts for biopharmaceutical development, yet the physiology of these cells types has yet to be fully established. Proteomics approaches can provide new insights into these cell types, aiding in recombinant protein production, cell growth regulation, and medium formulation. Optimization of sample preparations and protein database developments are enhancing the quantity and accuracy of proteomic results. In these ways, innovations in proteomics are enriching biotechnology and bioprocessing research across a wide spectrum of applications.

Introduction

There has been a transformation in biotechnology in the way that it now relies extensively on vast data sets generated using proteomics techniques. Both label free and comparative proteomics can now identify and quantify thousands of cellular proteins for therapeutic discovery purposes and to increase understanding of production hosts such as the Chinese hamster ovary (CHO) and Escherichia coli cells. At the earliest research stages, proteomics can identify new targets for infectious diseases, inflammatory diseases, chronic conditions, and cancer. Comparative proteomics between normal and diseased cells has the potential to identify differentially expressed proteins, which may represent novel targets and biomarkers. At the next stage in drug discovery, proteomics can be used to identify mAbs that might have therapeutic utility directly from the serum of patients (e.g., infectious disease). Finally, proteomics can be used to reveal differentially expressed host proteins between high and low biotherapeutic producers in order to aid cell line engineering efforts for increased recombinant protein productivity by manufacturing cells.

Proteomics aids disease target identification

Applying ‘omics tools to study the changes as cells progress from healthy to a diseased state is a valuable strategy for identifying potential therapeutic targets. In this role, comparative proteomics has been used for novel target identification in cancer, inflammatory, neurological and other diseases. For example, a comparison between human pulmonary adenocarcinoma and surrounding healthy tissue revealed over 30 differentially expressed proteins [1]. Two of the proteins significantly up-regulated in the cancer tissue were PKM2 and cofilin-1 [1]. On the basis of the proteomics results, PKM2 was targeted by RNA interference knockdown in vitro, which led to decreased cell growth and apoptosis induction in a cancer model cell line [1].

Protein interactions involved in disease progression can sometimes occur at the cell surface. Human cancer cell lines and osteoblast control cells were biotinylated to isolate surface proteins and then compared to identify over 150 significantly up-regulated proteins [2]. A surface proteomics experiment revealed that ephrin type-A receptor 2 (EphA2) is the most widely expressed surface protein on the cancer cells, suggesting it as a novel target for osteosarcoma biotherapeutics [2]. These cell surface receptors can then serve as targets for therapeutic intervention [2,3]. Proteomics analysis is useful for identification of biomarkers and novel drug targets as well as for evaluation of the physiological effects of drug candidates.

Proteomics discovers novel mAbs

MAb therapy has revolutionized the biotechnology industry and transformed the treatment of a range of different diseases. Proteomic methods are now being incorporated into the mAb discovery process as a complement to traditional hybridoma and phage display technologies. Both Cheung et al. [4] and Wine et al. [5•] applied proteomics to discover mAbs from the serum of immunized rabbit and mice. They employed affinity purification to the antibody mixture [4] and digested antibody fractions were then analyzed by nano-flow liquid chromatography coupled to tandem mass spectrometry (MS/MS). The spectra were then mapped to next generation sequencing (NGS) derived transcript sequences of the immunoglobulin heavy chain variable region to elucidate the antibody composition of a polyclonal serum response following immunization [5•]. In this way, proteomics along with other ‘omics tools, is now being used for the identification of potentially valuable mAbs directly from serum of animals.

Proteomics supports bioprocess development

Optimizing the manufacturing processes of biotherapeutics represents one avenue for making drug costs more affordable. Thus there is a desire to develop and utilize stable cell lines with high yields and even higher product quality. Proteomics can serve an important role in this effort by identifying those factors that enhance the cell’s capacity to produce high yields of protein therapeutics. CHO cell lines have been the dominant biotherapeutic production hosts due to their adaptability to bioprocessing and the presence of post-translational modifications compatible with humans. The recently published proteomic analysis of CHO cells [6•] has increased knowledge of CHO host cell proteins and pathways and will likely form the basis for future cell engineering efforts aimed at improving cell line characteristics. This large-scale analysis focused on intracellular proteins, the secretome and the glycoproteome, based on the original draft CHO genome [6•,7,8]. Pathway analysis revealed that protein processing and apoptosis related genes were enriched in expression, whereas steroid hormone and glycosphingolipid metabolic pathways were depleted [6•]. Comparative proteomics, such as isobaric tags for relative and absolute quantitation (iTRAQ), can be used to identify differentially expressed proteins associated with key cellular properties including protein production, cell growth, reduced apoptosis, favorable glycosylation, and optimized medium formulations (Figure 1).

Figure 1.

Overview of an optimized proteomics experiment for mammalian cell culture protein quantification. Proteins are extracted and subjected to reduction, alkylation, filter aided sample preparation, and digestion. Digested peptides are labeled with iTRAQ reagents, fractionated by bRPLC, and injected into LC/MS/MS. The resultant peaks are analyzed by mapping to CHO genome databases. Differences in peak levels represent relative protein expression levels between cell lines. Identification and quantification of proteins aids bioprocess development efforts to increase protein yields or other applications.

Proteomics differentiates growth conditions

Over the duration of cell culture, high cell density and accumulation of waste products contribute to apoptosis. In order to characterize a CHO IgG-9beta8 cell culture expressing an anti-Rhesus D factor monoclonal antibody at four time points, bioreactor samples were analyzed by an in-gel digestion coupled proteomics technique [9]. The goal of this study was to determine if different proteins are expressed as apoptosis is initiated as verified by a sharp increase in caspase 9 activity [9]. Results indicated that over time, there is an induction of the unfolded protein response that eventually leads to apoptosis [9]. Forty differentially expressed proteins were identified, including ones attributed to cytoskeletal organization, protein folding, and metabolism [9].

Delaying the onset of apoptosis can also increase protein production as demonstrated by coexpression of the anti-apoptosis gene Bcl-xL in CHO cells [10]. This approach was used in a related investigation of what changes take place in protein expression between exponential and stationary phases [11]. Furthermore, this study used in-solution digestion and comparative iTRAQ proteomics to identify 59 proteins with significantly different protein expression levels between the exponential and stationary phases, including binding immunoglobulin protein (BiP), protein disulfide isomerase (PDI), DNA replication licensing factors MCM2 and MCM5, transglutaminase-2, and clusterin [11]. After identifying and classifying differentially expressed proteins, it was observed that growth regulation and apoptotic proteins are highly expressed during the stationary phase [11]. This approach enabled the identification of a large number of total proteins and then confirmed significant changes in protein expression using Western blotting [11]. Results from this study have helped to identify dynamic protein changes that occur over time in a cell line coexpressing recombinant antibody and Bcl-xl, thus suggesting possible targets for prolonging the exponential phase.

Proteomics aids medium development

Methods that facilitate media formulation and development are highly desirable for an optimized and successful bioprocess. As a result, proteomics was used to identify proteins that aid in adaptation of a cell line from serum-bearing to serum-free media [12]. Expression of six proteins, including two molecular chaperones and four de novo nucleotide synthesis related proteins, was significantly increased in the serum-free cell culture [12]. Following the proteomics analysis, two of the chaperones identified (HSP60 and HSC70) were overexpressed, resulting in increased cell density up to 15% and decreased adaptation time by 33% [12].

In addition, proteomics has been used to identify components that can be added or withheld to improve media formulations that promote desirable growth and productivity characteristics. For example, secreted proteins from conditioned media samples were characterized in order to identify supplements that could be added to a serum-free media formulation to sustain cell growth [13]. Addition of identified growth factors, such as fibroblast growth factor 8, growth regulated alpha protein, hepatocyte growth factor, and macrophage colony stimulating factor 1, to a serum-free cloning media formulation led to increased cell density [13]. In another experiment, N-azido-galactosamine labeling was used to tag the mucin-type O-linked glycans of secreted proteins in order to enable their identification in cell-conditioned media [14]. The secretome of CHO-S and CHO DG44 cell lines were compared and 171 proteins were identified in both cell lines. In addition, 96 proteins were found to be unique to CHO DG44 and 85 proteins unique to CHO-S [14]. In another experiment, comparative proteomics was used to identify differences between cells cultivated in serum-free medium formulations with or without hydrolysates [15]. The changes in protein expression upon addition of hydrolysates, containing blends of peptides, free amino acids, vitamins, and trace elements, helped to explain the increased recombinant protein productivity in the hydro-lysate-containing media [15]. Proliferative proteins were up-regulated whereas pro-apoptotic proteins were down-regulated in the culture containing hydrolysates [15]. These studies demonstrate the capacity of proteomics to assist in the identification of novel media supplements and ultimately the optimization of the overall media formulation.

Proteomics identifies high productivity characteristics

In another report, the genetic factors contributing to high yields were studied using proteomics to compare various CHO-DHFR cell lines with different productivity levels under the same process conditions [16]. Results indicated a positive correlation between productivity and the expression of recombinant dihydrofolate reductase (DHFR), adaptor protein complex subunits, DNA repair proteins, and the ER translocation complex components [16]. Both transcriptomics and proteomics data were combined in order to improve understanding of the high producer phenotype [16]. In yet another study, high and low producing CHOK1SV cell lines generated using glutamine synthetase (GS) selection were compared by proteomics [17]. Results uncovered a group of 180 differentially expressed proteins, including 12 proteins associated with growth, metabolism, organization, and protein synthesis, which showed different expression levels over the duration of the cultures [17]. Another study of high and low producing cell lines derived from the same host CHO-K1 revealed that proteins involved in translation and folding were expressed at increased levels in high-producing cell lines [18].

Proteomics characterizes product quality attributes

Product quality including glycosylation modifications plays an important role in bioprocess development. Thus, proteomics may be useful for identifying factors that affect product glycosylation and especially sialylation. Comparison of various CHO cell lines identified different mutations in glycosylation pathways, suggesting that cell line differences may affect phenotypic properties such as product quality [8]. As revealed by the CHO cell line proteome, there are enrichments in glycosylation pathways at both the transcript and protein levels [6•]. For example, the highest percentage of proteins in the glycoproteome was N-acetylglucosaminyltransferases, which can affect the resulting product glycoforms [8].

Investigation of secreted proteins by click-chemistry based proteomics-methods elucidated proteins that are involved in the secretory pathway [14]. The method used N-azido-galactosamine labeling to tag mucin-type O-linked glycans on secreted proteins, thus enabling their identification by mass spectrometry [14]. This application of proteomics aids product quality improvement by identifying glycosylated proteins that help protein processing. Additionally, some host cell proteins (HCPs) are detrimental to product quality, posing problems such as purification, stability, and immunogenicity of the drug product [14]. Identification of critical HCPs serves to minimize drug product degradation after secretion.

Recent advancements in proteomics techniques for the CHO proteome

As the use of proteomics has become more widespread in the biotechnology industry, there have been efforts to enhance data collection and improve analytical tools. Multiple sample preparation techniques were coupled with two dimensional liquid chromatography techniques in order to identify over 6000 CHO proteins [6•]. Other methods were implemented in another study in order to maximize solubility and recovery of proteins with diverse physical and chemical properties. The optimal solubilizing factors for CHO cells were identified and implemented through a design of experiments approach [19].

Identification of secreted proteins can be challenging due to their low abundance. Protein recovery was optimized by varying chemical precipitants, concentration, and incubation duration for both gel-based and shotgun proteomics approaches [20••]. The results were used to optimize a method for identification of host cell proteins and over 170 proteins were identified [20••]. Further fine-tuning of proteomics methodologies has resulted in identification of the mitotic spindle proteins in CHO cells [21]. Isolation of the spindle aided in the discovery of factors affecting division and consequently, growth regulation [21].

In concert, CHO-specific proteomic databases have been implemented based on progressively improving sequence information for protein identification. Following the completed genome and proteome of the CHOK1 cell line, [6•,7], a large scale proteomic database for CHO cells, shown in Figure 2 (CHO Proteome Database; URL: http://chogenome.org/proteome.php), has been established.

Figure 2.

An example output for identified proteins in CHO-K1 proteome (CHO Proteome Database; URL: http://chogenome.org/proteomeSearch.php). (a) The protein name and accession number is retrieved after searching with a name. There are 6163 proteins total in this database (as shown by entering ‘%%’ for searching all the database proteins). (b) After clicking on an accession number, detailed information about a particular protein can be seen, including identified peptide sequences.

The detected proteins and their accession numbers are listed on the proteome search page of the website. The protein of interest can be searched either by protein name or accession number. After the proteins that meet the entered search criteria are listed, one can click on the accession numbers to get more detailed information. For each protein, the information, as shown in Figure 2b, contains general details such as its SwissProt annotation, GO annotation, and KEGG annotation and more specific information such as identified peptide sequences and false discovery rate (FDR) value. The presence of a CHO proteome database will facilitate integration of additional proteomics data and serve as a continually updated public resource for the CHO proteome going forward.

Recent advancements in proteomics techniques for the E. coli proteome

E. coli genome sequencing [22] has yielded opportunities for characterization of its transcriptome, proteome, inter-actome, metabolome, and physiome [23–27]. Although the first E. coli proteome was reported in the 1970s [28], recent advances in analytical technology such as two-dimensional gel electrophoresis (2-DE), MS, MS/MS [29,30], and combinations of chromatography and MS approaches have provided comparative studies. Roth-mann et al. [31] employed a five-step MudPIT experiment to identify the interactome of acyl carrier protein (ACP) in E. coli. Using their resin-based approach, the interaction between dihydrodipicolinate synthase and ACP has been shown to demonstrate ACP’s controlling role in bacterial cell wall formation.

Several groups have used the iTRAQ labeling technique for exploring the E. coli proteome [32–35]. Wu et al. [33] applied iTRAQ absolute quantitation for finding differentially regulated proteins in a recombinant strain of E. coli K4, which produces K4CPS, a biopolymer used in biotechnology industry [33]. Other research groups have used SILAC techniques to study the E. coli proteome [36,37,38•]. One group developed a novel SILAC medium (SILACE) for E. coli which provides faster growth and labeling efficiency as compared to E. coli grown in M9 medium, DMEM medium, and QconCAT medium [36]. Using the novel SILACE medium, they found more than 1200 proteins with significant changes in abundance for E. coli [36]. The pros and cons of three widely used methods are given in Table 1.

Table 1.

A comparison of proteomics techniques

Method	Pros	Cons
SILAC	Couples activity assay to expression profiling of live cells Offers greater coverage of proteome Diminishes error in quantitation compared to chemical labeling methods [39] Labels different amino acids residues Provides higher sensitivity than chemical labeling methods [39]	Causes difficulty in implementing analysis of certain samples such as tissues Leads to process delays in cell adaptation to dialyzed serum Mandates complex analysis involving simultaneous comparison of several samples Costs more compared to chemical labeling [39]
iTRAQ	Enables relative quantification in any cell type, as well as tissues and biological fluids Labels all peptides from tryptic proteolysis, resulting in comprehensive data [40] Allows for easy analysis of post translational modifications [41]	Requires longer MS and subsequent analysis time Necessitates small margin of error in sample preparation [42]
2-DE	Provides for simultaneous visualization of very large numbers of proteins Improves visualization through differential display format [43]	Is time consuming and sometimes semi-quantitative Renders limited dynamic range Creates problems for identifying hydrophobic proteins [43]

In conjunction with advanced proteomic methods, bioinformatics tools have been developed to aid in proteomics analysis of E. coli. Malinowska et al. [44] recently developed the bioinformatics software Diffprot, which implements a non-parametric approach for statistical analysis of MS-derived quantitative data which was successfully tested on E. coli iTRAQ experimental data.

Conclusions

The application of proteomics has advanced rapidly across multiple fields of research in order to identify new drugs and biomarkers, understand the mechanism of drug action, and discover potentially valuable mAbs. Recently, proteomics has been applied to improve bio-processing by increasing cell growth, minimizing cell death, and enhancing recombinant protein biotherapeutics production. As proteomics has evolved, more sensitive spectrometers and a number of comparative tools and technologies including iTRAQ and SILAC are being widely implemented to improve protein identification and quantitate changes in both mammalian and bacterial cell culture. In concert, more robust databases and bioinformatics tools are being introduced to assist in a more selective analysis of the exploding amount of data generated through proteomics analysis. Current and continual refinements in the tools and methodologies and the expansion of its use has ensured that proteomics will become an essential component of the biotechnology enterprise from discovery to manufacturing.